History is a long catalog of teams who marched in to save the day and, with gusto and paperwork, burned the day down instead. The Sack of Constantinople in 1204 might be the canonical entry. A crusading army meant to relieve the Holy Land stalled on logistics and debt, accepted a side quest, then"”nudged by Venetian interests and a confidence that God was surely aligned with their accounts receivable"”captured and looted the greatest Christian city of its age. If the brand promise was "defend Christendom," the shipped feature was "destroy its strongest bulwark." Own goal.
The drama reads medieval, but the mechanics are painfully modern: goal drift, perverse incentives, and the illusion that tactical wins add up to strategic victories. Builders see the same pattern when a roadmap is quietly re-optimized for quarterly optics or when a partner's milestone becomes the KPI that really matters. The city burns one decision at a time.
Consider Kodak's digital camera moment. The company literally invented the future and then protected its present by sidelining it. The act wasn't ignorance; it was fidelity to the current P&L and to a distribution machine tuned for film. The crusaders at Rochester marched under a familiar banner: save the core, fund the dividends. The result is in the history books.
Or New Coke"”an exquisitely executed market test that optimized for the wrong objective function. In blind trials, sweeter won against sips, so the company replatformed a cultural constant to nudge a small delta in preference curves. The backlash wasn't about taste; it was about trust and continuity. Constantinople burns again, this time with corn syrup.
What turns rescue into ruin?
- Mis-scoped mission: The Fourth Crusade started with a mission (liberate Jerusalem) and shipped a deliverable (pay off Venice, topple a claimant, sack a city). When teams declare a mission and then accept deliverables that don't ladder to it, entropy does the rest.
- Debt as compass: The army owed Venice; Venice had ships and a plan. Debts, whether financial or political, are strong magnets. In product orgs, calendar debt (promises to big customers) and architectural debt (commitments to an old platform) exert similar force. Pay attention to who you owe.
- Proxy metrics: Blind sip tests aren't a beverage; they're a sensor. Optimizing the sensor can de-optimize the system. Watch your proxies.
So how do you keep the city standing while you "save" it?
-
Bind incentives to mission, not milestones. Tie bonuses and roadmaps to durable outcomes: retention across cohorts, safety metrics, time-to-value for new segments. If a quarter trends toward shiny but shallow, the comp plan should feel wrong in the room.
-
Maintain an explicit "mission change log." If you accept a detour (a strategic partnership, a large bespoke deal, an acquisition), name the risk in writing and set a sunset. Crusades end; partnerships should too. If the detour keeps paying for its own extension, you might be Venice.
-
Manage proxy metrics like radioactive material. Proxies are fine"”indispensable"”but they need containers, exposure limits, and cross-checks. If you switch from sips to full cans, does the preference hold? If not, freeze rollout until you can reconcile the readings.
-
Demand counterfactuals ahead of force. "If we didn't do X, what would likely happen?" is not a rhetorical question. Before you sack a city"”or rewrite a product from scratch"”write the counterfactual path with data. Ship the cheaper path if it clears the mission bar.
-
Protect the bulwarks. The Byzantines' strength was layered defense and deep administration. Every modern system needs the equivalent: trusted identity, observability, safe defaults. When you mortgage the bulwarks for speed, remember that speed through ash is not speed.
Anecdote: On a late-stage enterprise deal years ago, we rushed a "one tiny exception" for a procurement portal that would "unlock the whole account." The exception required a session token to ride a custom header through an ancient proxy. Two weeks later, the same proxy stripped the header in production and we spent a weekend untangling phantom 401s while the champion watched momentum die in email. The "rescue" of the deal put the bulwarks (auth invariants) on pause. We shipped a patched tunnel; we also shipped a quiet reputation hit.
Another favorite: a DDoS "fix" that narrowed rate limits just enough to knock out our own webhooks during a partner launch. The attackers shrugged; the partner didn't. We had optimized the wrong boundary"”protecting the perimeter at the expense of the arteries. Every SOC has a Constantinople story.
And there are the strategic versions. Nokia's "we are a hardware company" story is told as a morality play about touchscreens. The deeper read is governance arithmetic: a successful org chart can defeat a promising product by scoring locally and losing globally. The crusaders didn't sack Byzantium in one meeting; they did it with a hundred plausible votes.
The cure is not cynicism. It's craft: clear missions, honest metrics, and a practice of checking whether a brilliant tactical move will burn the city you promised to defend. When in doubt, write it down. If you're about to switch targets because the shipwrights say the winds have changed, ask yourself who owns the fleet.
Finally, humor helps. The next time someone proposes a heroic pivot that will definitely save everything, ask"”deadpan"”"Is this more Jerusalem or more Constantinople?" If the room laughs, you're probably safe. If the room doesn't, post a guard on the treasury.